F-Droid Devs

80 Members
F-Droid development discussion only | Use #fdroid:f-droid.org for general, app- and repo-related matters | Meeting every Thursday at 11:30 UTC | This channel is publicly logged at https://matrix.f-droid.org/alias/%23fdroid-dev:f-droid.org 12 Servers

Load older messages

Timestamp Message
4 Jun 2020
20:04:00@bubu:bubu1.euBubuI get weird Could not resolve host: github.com errors when running too many jobs in parallel
20:09:28@freenode_cdesai:matrix.orgcdesaiI was going to say rate limits but this sounds like DNS
20:14:42@eighthave:matrix.org_hcI've seen weird stuff with too rapid DNS in gitlabci
20:18:45@bubu:bubu1.euBubuWell that's my local machine :_/
20:21:13@freenode_wb9688:matrix.orgwb9688 What DNS server do you use? I've never had any issues like that
20:21:36@freenode_cdesai:matrix.orgcdesaiyeah same and I use parallel to update a bunch of repos all the time (though probably not as many)
20:22:13@bubu:bubu1.euBubushitty isp
20:25:32@freenode_cdesai:matrix.orgcdesaicould use something that supports DoH/DoT
20:26:12@freenode_wb9688:matrix.orgwb9688 Why is DoH/DoT so hyped these days? What's wrong with plain old DNSCrypt?
20:28:14@freenode_wb9688:matrix.orgwb9688 Anyway, I have set up dnsdist locally to load balance between and and cache that. It's also set up to send queries for the .test TLD to my PowerDNS Authoritative Server. I haven't seen performance issues with this set up yet (unlike when I used dnsmasq lol), and it properly validates DNSSEC, plus it has DNSCrypt, DoT, and DoH support.
22:04:34@bubu:bubu1.euBubuwould gitlab be sad if I trigger 30 jobs in parallel?
22:20:16@eighthave:matrix.org_hcmaybe it makes more sense to trigger checkupdates per-appid whenever issuebot finds a related issue or merge request?
22:40:58@bubu:bubu1.euBubunot really
22:41:11@bubu:bubu1.euBubuor maybe as a debugging tool
22:41:40@bubu:bubu1.euBubuas in "this is what checkupdates looks for this appid"
22:41:58@bubu:bubu1.euBubuI've thought about that as well
23:09:29@freenode_izzy:matrix.orgizzy <_hc "I'll try 60 minutes for the build timeout"> Looks like that did it, thanks again!
23:56:42* @bubu:bubu1.euBubu wonders if we can spawn 3338 child-pipelines and be done with checkupdates in 5 minutes
5 Jun 2020
00:35:58@bubu:bubu1.euBubuthough 30 minutes is pretty good already
01:35:21@bubu:bubu1.euBubuthis should be working now. I can move that to one of the cfarm boxed but I don't get along with their ansible setup. If I can just dedicate gcc148 to this I can set this up.
01:36:29@bubu:bubu1.euBubu * this should be working now. I can move that to one of the cfarm boxed but I don't get along with their ansible setup. If I can just dedicate gcc148 to this I can set this up.
05:33:05@eighthave:matrix.org_hc Bubu: gitlab.com and our runners combined make no more than 100 parallel jobs.
05:33:36@eighthave:matrix.org_hc does checkupdates need the Android SDK? If not, you could use a much lighter image: which can take a couple minutes off the job
05:36:33@eighthave:matrix.org_hc like image: debian/stretch
05:39:15@eighthave:matrix.org_hcah, that job should not have the abillty to push straight to fdroid/fdroiddata git from gitlab-ci, those runners are not secure enough for that
05:39:45@eighthave:matrix.org_hcit can easily make a merge request, I have that working
05:41:37@eighthave:matrix.org_hcalso, this kind of big change should really be discussed in its own merge request, or an issue at least. this would not have been hard to set up in a test env
05:41:47@eighthave:matrix.org_hcI have that for the issuebot stuff
09:03:06@eighthave:matrix.org_hcI have to say, its cool that you got it working, I never would have thought it would have been possible in gitlab-ci.

There are no newer messages yet.

Back to Room List